Pr.9B.3

Local-­‐scale modeling of land surface processes in the cul5vated Sahel AMMA-­‐CATCH Niger C. VELLUET, J. DEMARTY, B. CAPPELAERE, I. BRAUD , N. BOULAIN, D. RAMIER, G. CHARVET, H. B.-­‐A. ISSOUFOU, M. BOUCHER, I. MAINASSARA, J.-­‐P. CHAZARIN, M. OÏ, H. YAHOU, N. BENARROSH, M. IBRAHIM, G. FAVREAU

MotivationsMaliWest AfricaNigerNigeriaSouth-­‐West Niger 100%01950 2000Local-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. Leduc et al., 2001 ; Cappelaere et al., 20091/11

Objective100%01950 2000 Evaluate the differences of the local processes on a fallowsavannah and a millet field with a process-based LSMLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 2/11

Tools: DataNWest Africa1 kmWankamacatchmentStudy area: Climate: South Sahel – semi-arid Rainfall: 400-600 mm/yr ETP: 2 300 mm/yr Soils: sandy soils prone to surface crusting Hydrology: endorheicLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 3/11

Tools: Simple Soil Plant Atmosphere modelSiSPATBraud et al., 1995Vertical (1D)AtmosphereVegetationSoilR NHGEvTrPRΔθWater and energyfluxes and storages:• Runoff (R)• Drainage (D)• Water storage (Δθ)• Evapotranspiration (ET)• Net Radiation (R N )• Sensible heat flux (H)• Soil heat flux (G)• Soil temperature (T)• Soil moisture (θ)DLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 4/11

Tools: Simple Soil Plant Atmosphere modelSiSPATBraud et al., 1995Vertical (1D)AtmosphereVegetationSoilR NHGEvTrPRΔθWater and energyfluxes and storages:• Runoff (R)• Drainage (D)• Water storage (Δθ)• Evapotranspiration (ET)• Net Radiation (R N )• Sensible heat flux (H)• Soil heat flux (G)• Soil temperature (T)• Soil moisture (θ)T,θDLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 4/11

ModellingSiSPATBraud et al., 1995Vertical (1D)AtmosphereVegetationSoil012070120400R NHEvTrH1: surface crustG H2 10 cmH3H4T,θH550 cm100 cm150 cm200 cm250 cmPRΔθWater and energyfluxes and storages:• Runoff (R)• Drainage (D)• Water storage (Δθ)• Evapotranspiration (ET)• Net Radiation (R N )• Sensible heat flux (H)• Soil heat flux (G)• Soil temperature (T)• Soil moisture (θ)Depth (cm)DObs (θ, T)Local-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 5/11

ModellingInitial conditions T i et θ iParameters:Soil and vegetation propertiesWater and energyfluxes and storages:• Runoff (R)• Drainage (D)• Water storage (Δθ)• Evapotranspiration (ET)• Net Radiation (R N )• Sensible heat flux (H)• Soil heat flux (G)• Soil temperature (T)• Soil moisture (θ)Local-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 5/11

ModellingInitial conditions T i et θ iMeteorological forcingWater and energyfluxes and storages:• Runoff (R)• Drainage (D)• Water storage (Δθ)• Evapotranspiration (ET)• Net Radiation (R N )• Sensible heat flux (H)• Soil heat flux (G)• Soil temperature (T)• Soil moisture (θ)Parameters:Soil and vegetation properties Calibration: 2 years (2005-2006) Validation: 3 years (2007-2009)Local-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 5/11

ModellingInitial conditions T i et θ iMeteorological forcingPhenology forcingLAI shrubFallowgrassWater and energyfluxes and storages:• Runoff (R)• Drainage (D)• Water storage (Δθ)• Evapotranspiration (ET)• Net Radiation (R N )• Sensible heat flux (H)• Soil heat flux (G)• Soil temperature (T)• Soil moisture (θ)MilletmilletshrubParameters:Soil and vegetation properties Calibration: 2 years (2005-2006) Validation: 3 years (2007-2009)Day of yearLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 5/11

ModellingInitial conditions T i et θ iMeteorological forcingPhenology forcingLAI shrubFallowgrassWater and energyfluxes and storages:• Runoff (R)• Drainage (D)• Water storage (Δθ)• Evapotranspiration (ET)• Net Radiation (R N )• Sensible heat flux (H)• Soil heat flux (G)• Soil temperature (T)• Soil moisture (θ)MilletGrazing milletParameters:Soil and vegetation propertiesshrub Calibration: 2 years (2005-2006) Validation: 3 years (2007-2009)Day of yearLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 5/11

Calibration: Hydrodynamic propertiesθ = θ +rθ⎡⎡ ⎛⎛⎢⎢1+ ⎜⎜⎢⎢⎣⎣⎝⎝sathh−θg⎞⎞⎟⎟⎠⎠rn⎤⎤⎥⎥⎥⎥⎦⎦1−2/nRetention capacity:van Genuchten (1980)Hydraulic conductivity:Brooks & Corey (1964)( θ )h g n θ sat K sat βm - m 3 /m 3 m/s -K=K sat⎛⎛ θθ ⎟⎟ ⎟⎟ ⎞⎞⎜⎜⎜⎜⎝⎝ sat ⎠⎠β6/11

Calibration: Hydrodynamic propertiesθ = θ +rθ⎡⎡ ⎛⎛⎢⎢1+ ⎜⎜⎢⎢⎣⎣⎝⎝Fallow savannahH1: 0-1 cm -1,20h g n θ sat K sat βm - m 3 /m 3 m/s -H2: 1-20 cm -0,50 2,79.10 -6H3: 20-70 cm0,309.10 -8 6,05.10 -6H4: 70-120 cm-0,289.10 -62,6H5: 120-400 cm 3.10 -5Millet fieldsathhH1: 0-1 cm3.10 -7-0,500,29H2: 1-20 cm 2,72,5.10 -6H3: 20-70 cm−θg⎞⎞⎟⎟⎠⎠rn⎤⎤⎥⎥⎥⎥⎦⎦1−2/nRetention capacity:van Genuchten (1980)Hydraulic conductivity:Brooks & Corey (1964)H4: 70-120 cm-0,30 0,30 2.10 -52,6H5: 120-400 cm 1.10 -5K( θ )=5,06,01.10 -5 5,55,0K sat⎛⎛ θθ ⎟⎟ ⎟⎟ ⎞⎞⎜⎜⎜⎜⎝⎝ sat ⎠⎠β6/11

Calibration: Hydrodynamic propertiesθ = θ +rθ⎡⎡ ⎛⎛⎢⎢1+ ⎜⎜⎢⎢⎣⎣⎝⎝sathh−θg⎞⎞⎟⎟⎠⎠Fallow savannahrn⎤⎤⎥⎥⎥⎥⎦⎦1−2/nH1: 0-1 cm -1,20Retention capacity:van Genuchten (1980)Hydraulic conductivity:Brooks & Corey (1964)K( θ )h g n θ sat K sat βm - m 3 /m 3 m/s -H2: 1-20 cm -0,50 2,79.10 -6H3: 20-70 cm0,30=9.10 -8 6,05.10 -6H4: 70-120 cm-0,289.10 -62,6H5: 120-400 cm 3.10 -5Millet fieldH1: 0-1 cm3.10 -7-0,500,29H2: 1-20 cm 2,72,5.10 -6H3: 20-70 cmH4: 70-120 cm-0,30 0,30 2.10 -52,6H5: 120-400 cm 1.10 -55,06,01.10 -5 5,55,0K sat⎛⎛ θθ ⎟⎟ ⎟⎟ ⎞⎞⎜⎜⎜⎜⎝⎝ sat ⎠⎠β6/11

Calibration: Hydrodynamic propertiesθ = θ +rθ⎡⎡ ⎛⎛⎢⎢1+ ⎜⎜⎢⎢⎣⎣⎝⎝sathh−θg⎞⎞⎟⎟⎠⎠Fallow savannahrn⎤⎤⎥⎥⎥⎥⎦⎦1−2/nH1: 0-1 cm -1,20Retention capacity:van Genuchten (1980)Hydraulic conductivity:Brooks & Corey (1964)K( θ )h g n θ sat K sat βm - m 3 /m 3 m/s -H2: 1-20 cm -0,50 2,79.10 -6H3: 20-70 cm0,30=9.10 -8 6,05.10 -6H4: 70-120 cm-0,289.10 -62,6H5: 120-400 cm 3.10 -5Millet fieldH1: 0-1 cm -0,50H2: 1-20 cm 0,29 2,5.10 -6H3: 20-70 cmH4: 70-120 cm-0,30 0,30 2.10 -52,6H5: 120-400 cm 1.10 -52,75,03.10 -7 6,01.10 -5 5,55,0K sat⎛⎛ θθ ⎟⎟ ⎟⎟ ⎞⎞⎜⎜⎜⎜⎝⎝ sat ⎠⎠β6/11

Calibration: Hydrodynamic propertiesθ = θ +rθ⎡⎡ ⎛⎛⎢⎢1+ ⎜⎜⎢⎢⎣⎣⎝⎝sathh−θg⎞⎞⎟⎟⎠⎠Fallow savannahrn⎤⎤⎥⎥⎥⎥⎦⎦1−2/nH1: 0-1 cm -1,20Retention capacity:van Genuchten (1980)Hydraulic conductivity:Brooks & Corey (1964)K( θ )h g n θ sat K sat βm - m 3 /m 3 m/s -H2: 1-20 cm -0,50 2,79.10 -6H3: 20-70 cm0,30=9.10 -8 6,05.10 -6H4: 70-120 cm-0,289.10 -62,6H5: 120-400 cm 3.10 -5Millet fieldH1: 0-1 cm3.10 -7-0,500,29H2: 1-20 cm 2,72,5.10 -6H3: 20-70 cmH4: 70-120 cm-0,30 0,30 2.10 -52,6H5: 120-400 cm 1.10 -55,06,01.10 -5 5,55,0K sat⎛⎛ θθ ⎟⎟ ⎟⎟ ⎞⎞⎜⎜⎜⎜⎝⎝ sat ⎠⎠ Parameters: • Physically realistic• Similar for the both sites, except at surface• Surface crust stronger on fallow 6/11β

Calibration-Validation: Soil moistureθ soil (m 3 .m -3 ) Fallow savannahMillet field2005 2006 2007 2008 200910 cm2005 2006 2007 2008 200950 cm150 cm250 cmModelObsObs uncertaintyDay of year Model Performance:• Satisfactory for both sites, wet & dry years• Similar for calibration & validation• Better for fallow site Highlights:• Fallow: little infiltration below 1m• Millet: infiltration through the whole profileLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 7/11

Calibration-Validation: Energy fluxes800Fallow savannah2005 2006 2007 2008 2009R N Model (W/m²)Millet field2005 2006 2007 2008 20090-2008000-2008000-200800G H ET 0-2000 800 0 800 0 800 0 800 0 800 0 800 0 800 0 800 0 800 0 800Observations (W/m²)Local-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 8/11

Calibration-Validation: Energy fluxesModel (W/m²) Fallow savannahMillet field2005 2006 2007 2008 2009 2005 2006 2007 2008 2009800R N RMSE: 12 – 20 Bias: 6 – 10 RMSE: 16 – 21 Bias: -5 – 50-2008000-2008000-2008000-200G RMSE: 17 – 27 Bias: -7 – -5 RMSE: 16 – 21 Bias: -5 – 5H RMSE: 28 – 44 Bias: -9 – -2 RMSE: 24 – 30 Bias: -7 – 1ET RMSE: 34 – 47 Bias: -5 – 3 RMSE: 25 – 30 Bias: -5 – 40 800 0 800 0 800 0 800 0 800 0 800 0 800 0 800 0 800 0 800Observations (W/m²) Model Performance:• Satisfactory for both sites, wet & dry years• Similar for calibration & validation• Very good for R N• Good for G• Better H & ET for millet siteLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 8/11

Results: Annual water budget600600500500400400300200P30020010010000-1002005 2006 2007 2008 2009-1002005 2006 2007 2008 2009 Rainfall variability: P=420–580 mm/yr Evapotranspiration (ET):Fallow: 80–95% of PMillet: 85–110% of P Evaporation vs.Transpiration:Ev>TrFallow: T = 20–80% of ETMillet: T = of ET Runoff (R):Fallow: 5–20% of PMillet: 2–15% of P Drainage (D):Fallow: no drainageMillet: 0-5% of rainfall GW recharge Water stock (Δθ):Fallow: no interannual storage Δθ=0mmMillet: Δθ=9/11

Results: Annual water budgetFallow savannah6005004003002001000-1002005 2006 2007 2008 2009PET6005004003002001000-100Millet field2005 2006 2007 2008 2009 Rainfall variability: P=420–580 mm/yr Evapotranspiration (ET):Fallow: 80–95% of PMillet: 85–110% of P Evaporation vs.Transpiration:Ev>TrFallow: T = 20–80% of ETMillet: T = of ET Runoff (R):Fallow: 5–20% of PMillet: 2–15% of P Drainage (D):Fallow: no drainageMillet: 0-5% of rainfall GW recharge Water stock (Δθ):Fallow: no interannual storage Δθ=0mmMillet: Δθ=9/11

Results: Annual water budgetFallow savannah6005004003002001000-1002005 2006 2007 2008 2009PTEv6005004003002001000-100Millet field2005 2006 2007 2008 2009 Rainfall variability: P=420–580 mm/yr Evapotranspiration (ET):Fallow: 80–95% of PMillet: 85–110% of P Evaporation vs.Transpiration:Ev>TrFallow: T = 20–45% of ETMillet: T = 40–47% of ET Runoff (R):Fallow: 5–20% of PMillet: 2–15% of P Drainage (D):Fallow: no drainageMillet: 0-5% of rainfall GW recharge Water stock (Δθ):Fallow: no interannual storage Δθ=0mmMillet: Δθ=9/11

Results: Annual water budgetFallow savannah6005004003002001000-1002005 2006 2007 2008 2009PRTEv6005004003002001000-100Millet field2005 2006 2007 2008 2009 Rainfall variability: P=420–580 mm/yr Evapotranspiration (ET):Fallow: 80–95% of PMillet: 85–110% of P Evaporation vs.Transpiration:Ev>TrFallow: T = 20–45% of ETMillet: T = 40–47% of ET Runoff (R):Fallow: 5–20% of PMillet: 2–15% of P Drainage (D):Fallow: no drainageMillet: 0-5% of rainfall GW recharge Water stock (Δθ):Fallow: no interannual storage Δθ=0mmMillet: Δθ=9/11

Results: Annual water budgetFallow savannah6005004003002001000-1002005 2006 2007 2008 2009PDRTEv6005004003002001000-100Millet field2005 2006 2007 2008 2009 Rainfall variability: P=420–580 mm/yr Evapotranspiration (ET):Fallow: 80–95% of PMillet: 85–110% of P Evaporation vs.Transpiration:Ev>TrFallow: T = 20–45% of ETMillet: T = 40–47% of ET Runoff (R):Fallow: 5–20% of PMillet: 2–15% of P Drainage (D):Fallow: no drainageMillet: 0-5% of P GW recharge Water stock (Δθ):Fallow: no interannual storage Δθ=0mmMillet: Δθ=9/11

Results: Annual water budgetFallow savannah6005004003002001000-1002005 2006 2007 2008 2009PΔθDRTEv6005004003002001000-100Millet field2005 2006 2007 2008 2009 Rainfall variability: P=420–580 mm/yr Evapotranspiration (ET):Fallow: 80–95% of PMillet: 85–110% of P Evaporation vs.Transpiration:Ev>TrFallow: T = 20–45% of ETMillet: T = 40–47% of ET Runoff (R):Fallow: 5–20% of PMillet: 2–15% of P Drainage (D):Fallow: no drainageMillet: 0-5% of P GW recharge Water storage (Δθ):Fallow: no interannual storage Δθ=0mmMillet: Δθ = -25–+63 mm/yr9/11

Results: Monthly water budgetmm/month Fallow savannahRunoff Evaporation TranspirationMillet fieldRunoff Evaporation TranspirationLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 10/11

Results: Monthly water budgetmm/month Fallow savannahRunoff Evaporation TranspirationMillet fieldRunoff Evaporation Transpiration Inter-annual variability stronger on the fallowLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 10/11

Results: Monthly water budgetmm/month Fallow savannahRunoff Evaporation TranspirationMillet fieldRunoff Evaporation Transpiration Inter-annual variability stronger on the fallowLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 10/11

Results: Monthly water budgetmm/month Fallow savannahRunoff Evaporation TranspirationMillet fieldRunoff Evaporation Transpiration Inter-annual variability stronger on the fallow Seasonal distribution different for the 2 sitesLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 10/11

Conclusions Evaluate the differences of the local processes on a fallow savannah and amillet field with a process-based model Model performance: calibration & validation, satisfactory/stable for wet & dryyears, millet & fallow Process-based modeling complete, detailed water & energy budgets, forvariable meteorological and vegetation conditions Comparison of the millet & fallow water balances (2005-2006 - Ramier etal., 2009) main conclusions confirmed Inter-annual variability comparison of land cover types varies with yearsPerspectives Longer periods to account for the marked variability of the Sahel climate Effects of land management practices (agro forestry, grazing, …) ? Impact of uncertainties on inputs and parameters ?Local-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 11/11

Thank you for your aRen5on ! C. VELLUET, J. DEMARTY, B. CAPPELAERE, I. BRAUD , N. BOULAIN, D. RAMIER, G. CHARVET, H. B.-­‐A. ISSOUFOU, M. BOUCHER, I. MAINASSARA, J.-­‐P. CHAZARIN, M. OÏ, H. YAHOU, N. BENARROSH, M. IBRAHIM, G. FAVREAU

Soil temperatures2006 Fallow savannah 50 cm 150 cm Obs Model T(°C) RMSE : 2.3 Biais : -­‐2.2 Corr : 0.98 RMSE : 1.5 Bias : -­‐1.4 Corr : 0.98 2007 T(°C) RMSE : 2.1 Biais : -­‐2.0 Corr : 0.99 RMSE : 1.4 Bias : -­‐1.3 Corr : 0.98 2008 T(°C) RMSE : 1.8 Bias : -­‐1.7 Corr : 0.98 RMSE : 1.1 Bias : -­‐1.0 Corr : 0.98 2009 T(°C) RMSE : 1.5 Bias : -­‐1.4 Corr : 0.99 RMSE : 1.0 Bias : -­‐0.9 Corr : 0.98 Day of year

Soil temperatures2006 Millet field 50 cm 150 cm Obs Model T(°C) RMSE : 0.9 Bias : -­‐0.7 Corr : 0.97 RMSE : 0.4 Bias : -­‐0.2 Corr : 0.98 2007 T(°C) RMSE : 0.7 Bias : -­‐0.3 Corr : 0.98 RMSE : 0.5 Bias : 0.2 Corr : 0.98 2008 T(°C) RMSE : 0.9 Bias : 0 Corr : 0.97 RMSE : 0.8 Bias : 0.5 Corr : 0.96 2009 T(°C) RMSE : 1.1 Bias : -­‐0.2 Corr : 0.95 RMSE : 0.9 Bias : 0.3 Corr : 0.92 Day of year

ParametersOther parameters Units Fallow savannah Millet fieldVegetation albedo - 0,24 0,24Soil albedo (dry/wet) - 0,20/0,33 0,20/0,34Thermal conductivity J/m 3 /K 1,20.10 6 1,20.10 6Minimal stomatal resitance s/m 80 120Total plant resistance s/m 6,50.10 12 6,50.10 12

Soil granulometry and density Fallow savannah H1 : 0-1 cmH2 : 1-20 cmH3 : 20-70 cmH4 : 70-120 cmH5 : 120-400 cmSilt Clay Sand Density Obs depth Model nodes

Soil granulometry and density Millet field H1 : 0-1 cmH2 : 1-20 cmH3 : 20-70 cmH4 : 70-120 cmH5 : 120-400 cmSilt Clay Sand Density Obs depth Model nodes

Calibration-Validation: Energy fluxes on both sitesModel (W/m²) Fallow savannah2005 2006 2007 2008 2009800R N RMSE (Bias)Millet field2005 2006 2007 2008 20090-200800G 12 (6)20 (10)15 (9)17 (8)17 (6)19 (5)16 (0)19 (-5)21 (-6)18 (-7)0-200800H 27 (-7)17 (-4)19 (-5)25 (-6)21 (-5)19 (-6)23 (-3)17 (-4)19 (-7)17 (-4)0-200800ET 44 (-2)37 (-9)31 (-6)29 (-4)28 (-2)28 (-1)26 (-7)26 (-6)24 (-2)30 (1)0-20043 (-5)47 (3)34 (-5)40 (-3)34 (-3)30 (-3)0 800 0 800 0 800 0 800 0 800 0 800 0 800 0 800 0 800 0 800Observations (W/m²)26 (4)25 (2)27 (-2)27 (-5)Local-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 8/13

Results: Seasonal energy budgetW/m² Fallow savannahDry season (January-May)Wet season (June-October)Millet fieldDry season (January-May)Wet season (June-October) Interannualpartitionning of H-LE:more variable on fallow Rn higher on thefallowLocal-­‐scale modeling of land surface processes in the cul5vated Sahel – Velluet et al. 12/13